Dear Graham, Best would be to create alternative residues for the two proteins. A could be protein 1, B protein 2. As far as refinement programs are concerned, the alternatives need not to be identical and both alternatives are correctly linked to the common chain. You may have to do some pdb-hacking with an editor though. A quick test with coot revealed that the "mutate residue" option changes both alternatives.
Since you mention that both proteins are extremely similar, I would consider creating a consensus sequence for them, like was done for e.g. the antibodies or the serine proteases. This makes description and discussion of the different residues in papers a lot easier. You may get some trouble with some refinement programs, because some programs are no longer able to handle insertions and deletions. However, these are allowed by the pdb definitions and I consider not handling them a programming oversight. Best, Herman -----Ursprüngliche Nachricht----- Von: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Graham Robinson Gesendet: Mittwoch, 11. Januar 2017 17:37 An: CCP4BB@JISCMAIL.AC.UK Betreff: Re: [ccp4bb] Averaging of different proteins in heteromeric complex Dear Crystallographers, Thank you very much Andrew, David, and Jacob (as well as others outside of the BB) for very helpful input on this problem. It does in fact appear that I have a very nice, statistically disordered structure. In published structures containing statistical disorder authors have placed the different proteins in density according to knowledge from previous structures, and assigned occupancy of 0.5 to the entire chain. This is not possible in my case: I have a large, heteromeric ring, and one of the reasons for solving the structure in the first place was to characterise the arrangement of the two different proteins in the ring. Instead, I would like to build a consensus model, which represents both proteins. The resolved regions of each protein are extremely similar in primary sequence, and so, in the positions where the amino acids differ, I intend to prune the side chain either to the last common atom or to Ala, and name the residue a generic name (DLR, for DuaL Residue for example). I'd be interested to know if this a reasonable approach, and whether others have done a similar thing before? The N-termini of the two proteins (not resolved) differ meaning that while the resolved regions are very similar in sequence the chain numbering differs. Is there a sensible way to account for this in a consensus chain? Is it reasonable to adopt chain numbering of one of the two proteins? Thank you very much, Graham
